Sas version 8.2 or higher

All statistical analyses were performed in those who completed the study using SAS version 8.2 or higher (SAS Institute, Cary, NC, USA). Participants without a Week 52 HbA1c or weight value were treated as failures for the composite endpoints. All statistical tests were conducted at a 5% significance level.
The composite endpoints were analyzed overall, irrespective of baseline HbA1c, using logistic regression models with a factor for treatment (alogliptin or glipizide) and with baseline HbA1c and baseline weight as covariates. Additional logistic regression models were fitted separately for patients with a baseline HbA1c <8.0% and ≥8.0%.

Values for the pharmacokinetic parameters were estimated by noncompartmental methods using Phoenix WinNonlin, version 6.1 (Pharsight Corporation, Mountain View, California) or SAS version 8.2 or higher (SAS Institute, Inc., Cary, North Carolina). Pharmacokinetic parameters included the maximum observed plasma concentration (C_max), area under the plasma concentration‐time curve (AUC) during a dosing interval (AUC₂₄ for once‐daily administration, AUC₁₂ for twice‐daily administration, AUC_t for time zero to the time of the last measurable concentration, or AUC_∞ for time zero to infinity, as appropriate), and T_max. Additional pharmacokinetic parameters included t_½ and trough concentration (C_trough).